PiCo: Active Manifold Canonicalization for Robust Robotic Visual Anomaly Detection

2026-03-24

AI Summary

This work addresses the challenge of visual anomaly detection for robotics under complex operational conditions—such as varying 6-DoF poses, illumination changes, and shadows—where semantic anomalies and physical disturbances are tightly coupled. To tackle this, the authors propose an active manifold normalization paradigm that actively maps observations onto a condition-invariant canonical manifold through a cascaded mechanism. First, the robot actively reorients objects to reduce geometric uncertainty; then, a three-stage denoising architecture jointly suppresses multi-scale interference at the input, feature, and semantic levels. This study pioneers a shift from passive feature learning to an active physical-neural co-normalization framework, achieving state-of-the-art performance on the M2AD benchmark with 93.7% O-AUROC in static scenes and 98.5% accuracy in active closed-loop settings, substantially enhancing detection robustness and efficacy in industrial applications.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Adversarial Learning & Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Industrial deployment of robotic visual anomaly detection (VAD) is fundamentally constrained by passive perception under diverse 6-DoF pose configurations and unstable operating conditions such as illumination changes and shadows, where intrinsic semantic anomalies and physical disturbances coexist and interact. To overcome these limitations, a paradigm shift from passive feature learning to Active Canonicalization is proposed. PiCo (Pose-in-Condition Canonicalization) is introduced as a unified framework that actively projects observations onto a condition-invariant canonical manifold. PiCo operates through a cascaded mechanism. The first stage, Active Physical Canonicalization, enables a robotic agent to reorient objects in order to reduce geometric uncertainty at its source. The second stage, Neural Latent Canonicalization, adopts a three-stage denoising hierarchy consisting of photometric processing at the input level, latent refinement at the feature level, and contextual reasoning at the semantic level, progressively eliminating nuisance factors across representational scales. Extensive evaluations on the large-scale M2AD benchmark demonstrate the superiority of this paradigm. PiCo achieves a state-of-the-art 93.7% O-AUROC, representing a 3.7% improvement over prior methods in static settings, and attains 98.5% accuracy in active closed-loop scenarios. These results demonstrate that active manifold canonicalization is critical for robust embodied perception.

Problem

• visual anomaly detection
• 6-DoF pose
• illumination changes
• physical disturbances
• semantic anomalies

Innovation

• Active Canonicalization
• Pose-in-Condition Canonicalization
• Robotic Visual Anomaly Detection
• Manifold Learning
• Embodied Perception